Home Run Update – August 31
As the reader should know, it has been an interesting story of home run hitting during the Statcast era. There was a steady rise in home run hitting from 2015 through 2017 with total HR counts of 4909, 5610, and 6105, respectively. Then the home run count in the 2018 season dropped to 5585, and then the count dramatically increased to 6776 in the 2019 season.
Given that we have seen five months of baseball in the current 2021 season, we have a good idea of what is happening with respect to home run hitting this season. In this post I’ll focus on 2021 changes in values of the hitter launch conditions, specifically launch angle and launch velocity, and how these launch conditions have impacted home run hitting.
Home Runs and Launch Conditions
Here is a graph of the launch variables (launch angle and launch speed) for a sample of batted balls in the Statcast period. The red points correspond to home runs. I’ve drawn a black box where the launch angle is between 20 and 40 degrees and the exit velocity is between 95 and 115 mph — this is the region where practically all of the home runs fall.
Launch Conditions on Balls in Play
We focus on the black box area. We divide both the range of launch angle and launch velocity values into four subintervals, so the black box area is divided into 16 subregions. For each of the four recent full seasons (2017, 2018, 2019, 2021), we display the percentage of batted balls in each of the subregions in the following graph. For example, looking at the 2017 Season graph (top left panel), we see that 0.7% of all batted balls had launch angles between 35 and 40 degrees and launch speeds between 95 and 100 MPG.
Generally, batters are currently hitting a higher rate of balls in these home-run friendly launch variable regions. For example, look at the rows of the 2021 table (labeled in orange) where the launch velocity is between 100 and 110 mph. If you look across all seasons, we see the percentages in these two rows have steadily climbed from 2017 to 2021. If the launch conditions were the major input for predicting home runs, one would expect the home run rate would be the highest in 2021.
Home Run Rates – Playing with the 2018 Ball?
We know that the rate of hitting home runs has not steadily increased over the 2017-2021 time period. There is clear evidence that the properties of the ball have not stayed constant over this period. This is demonstrated in the following graph that displays the home run rate (in percentage) for each of the subregions of the (launch angle, launch velocity) space. Let’s focus on the bottom two rows (colored in orange) corresponding to launch velocity values between 95 and 105 mph. Looking across the 2017-2021 period, we see that the home run rates in these regions are respectively HIGH, LOW, HIGH, and LOW. In fact, the drag characteristics of the 2021 ball appear to be higher than the 2019 ball and approximately the same as the 2018 ball.
How Many Home Runs in 2021?
Currently (through games of August 29) we have seen 4740 home runs. Also we’ve played a total of 3912 out of a possible 4860 games. Assuming that we’ll continue to see home runs at the current rate, a simple estimate at the total home run count is:
hr2021 <- 4740 games2021 <- 3912 (prediction <- hr2021 * (4860 / games2021))  5888.65
It would be easy to construct a model that would allow one to construct a 90% prediction interval for the total HR count. Based on my experience, I am confident that the total 2021 home run count will be between 5800 and 6000. (Significantly lower than the 2019 HR count of 6776.). Here is an older post where I illustrate the use of a random effects model in predicting home runs.
- Two Forces. It is interesting that there are two forces that significantly impact home run production. Players are indeed putting more balls in play with higher exit velocities and “home-run friendly” launch angles which should increase home run production. But the relatively dead baseball (with the high drag) is causing home run production to decrease in the 2021 season. And we predict about 800 fewer home runs hit in 2021 compared to 2019.
- R Code? All of the R code for this exercise can be found at my Github Gist site. This includes reading in the Statcast data from my Github site.
- Shiny App. I also have a live Shiny app at https://bayesball.shinyapps.io/HomeRunCompare/ that allows one to compare the batted ball or home run rates over the launch conditions space for any two seasons of interest. By doing a number of comparisons using this app, that will reinforce the conclusions from this brief study.